Health
ChatGPT Tested for Medical Triage Accuracy in Structured Clinical Study
A new study published in Nature Medicine evaluates ChatGPT's performance in making triage recommendations using structured clinical scenarios. The findings offer one of the most rigorous assessments yet of whether large language models can reliably assist in medical decision-making.
Key Takeaways
- A Nature Medicine study tested ChatGPT's ability to make medical triage recommendations using standardized clinical scenarios
- The model performed well on clearly critical cases but struggled with nuanced middle-acuity decisions
- Inconsistency in repeated evaluations raises reliability concerns for clinical deployment
- Researchers recommend human-AI collaboration rather than autonomous AI triage systems
DE
DT Editorial AI··via nature.com